Computed tomography has revolutionized the practice of medicine and the role of imaging in health care. The imaging technology that was first successfully implemented nearly 35 years ago through the work of British engineer Sir Godfrey Hounsfield has shown marked evolution, bearing little resemblance to the capabilities of its predecessors, yet retaining many of the original concepts in the design. This article outlines the history of the development of CT, including the challenges that were overcome to allow for multidetector CT (MDCT), the products that are currently available in the market, and the future directions of CT.

A user-friendly scanner interface and workstation is the hallmark of Philips Medical?s Brilliance scanner.

The Evolutionary Biology of MDCT

The development of any technology can be characterized by incremental innovation punctuated by epochs of radical innovation. The perceived inadequacies of a certain component of a design often pave the way for improvement. But improvement is rarely brought about without additional challenges and some compromise. The final (or current) design may bear more resemblance to its more remote than proximate ancestors.

The timeline of CT has a similar story. First- and second-generation scanners used a rotate-translate principle—ie, to acquire one axial image, the x-ray source/detector had to move both circumferentially and linearly to get all desired projections. First-generation scanners had the highest scatter rejection rate due to their use of parallel ray geometry. Subsequent generations have used a fan/cone beam, which increases coverage at the expense of increased scatter. The use of an increased number of detectors along the axial plane (and broadening of the fan beam) obviated the need for translation in the axial plane, leading to third-generation scanner (rotate-rotate) technology. The fourth-generation scanner (rotate-stationary) corrected for the ring artifacts seen in third-generation systems, due to electronic drift in their detectors. However, with improved detector electronics and preprocessing, the rotate-stationary paradigm was short-lived, and current scanners follow the rotate-rotate principle. Fifth-generation scanners (electron-beam CT) were designed specifically for cardiac imaging and achieved a high temporal resolution (50 ms) as there was no moving component to the system (stationary-stationary). The technology lacked sufficient resolution to find widespread use, but it was a major player in determining the prognostic significance of coronary artery calcium.

Unquestionably, the radical innovation in CT (sixth-generation helical CT scanner technology) was implementation of the slip-ring interconnect, which allowed the stationary elements of the CT to be divorced from the moving component, thus permitting continuous rotation of the gantry. This allowed volume/helical coverage of the body part in a shorter period of time, such as a single breath-hold, and permitted reconstruction of data along any point of the z-axis—paving the way for 3D techniques.

MDCT (seventh-generation) facilitates the goals of helical CT in terms of volume coverage by increasing the coverage along the z-axis (longitudinal or the axis along which the patient is being scanned) per gantry rotation. An increase in detector number by a factor “N” leads to:

GE Healthcare?s VCT has a high dose efficiency, as the focal spot matches the detector.
  1. increased speed for the same coverage by a factor N (shorter breath-hold, less volume of iodinated contrast medium);
  2. increased coverage for the same speed (increased field of view in oncological, trauma, and peripheral vascular imaging);
  3. thinner collimations by a factor N for the same speed and field of view (CT angiography, isotropic imaging); or
  4. any combination of 1, 2, and 3—or put another way, isotropic submillimeter volume CT in a breath-hold at the lowest possible dose.

Behind the Scenes

The progress in MDCT has not been merely a numeric challenge leading to an increasing number of detectors. The acronym MDCT refers to only one aspect of CT, which is the volume coverage afforded by the multiplicity of detectors aligned in the z-axis. Focused admiration of this term does injustice to several components that have evolved in concert to make “isotropic submillimeter volume CT in a breath-hold at the lowest possible dose” a clinical reality. In fact, CT represents a marriage of such diverse areas of science and technology as computational power, detector science, x-ray tube/gantry design, and complex mathematical algorithms.

CT is, by definition, a union of computers and radiography. It is no surprise, therefore, that advancement in computers is the most important determinant of the progress of CT. First-generation scanners spent minutes reconstructing data from 28,800 rays (ie, 160 rays for 180 projections). Today, algebraic data from 800,000 data points (800 rays for 1,000 projections) can be computed in a fraction of a second. Increasing computational power is just as important in allowing the radiologist to view the large number of images or 3D reformatting at the workstation or a remote thin client, and in the storage and transmission of data to the local PACS/RIS. Indeed, any advancement in CT that results in more images (such as the ultrahigh spatial resolution of flat-panel detectors) will inspire further computational challenges.

Essentially, two assumptions are made in helical MDCT. The first is that the raw data has been obtained in a coplanar (axial) configuration. It has not; it has been obtained in a helical trajectory. (Helix is what you get when you combine rotation with concomitant linear motion.) To correct for this, interpolation is performed. When the helical pitch is greater than 1 (ie, the table moves a distance greater than the nominal beam width in one gantry rotation), interpolation increases the effective slice width (the slice width is determined by the detector width). Z?filtering is a novel spiral interpolation technique (such as the adaptive axial interpolation used by Siemens and the MUSCOT algorithm used by Toshiba) that corrects for the broadening of the slice width (and, therefore, degradation of longitudinal resolution) seen with imaging at a pitch greater than 1.

The second assumption is that the rays that result in the image travel at a right angle to the patient’s longitudinal axis. It does for a single-detector CT. This principle is violated in MDCT proportionate to the number of detectors. The angle made by the rays with the center plane is known as the “cone angle.” The cone angle can be ignored in up to 4 detectors, beyond which artifacts result (particularly at high-contrast objects and in the outer detectors), and the cone angle needs to be accounted for by algorithms, such as 3D back projection (Toshiba America Medical Systems, Tustin, Calif; and Philips Medical Systems, Andover, Mass), weighted hyperplane reconstructions (GE Healthcare, Waukesha, Wis), and adaptive multiplane reconstruction (Siemens Medical Solutions, Malvern, Pa).

Another advance in CT has been faster gantry rotation times. This has been particularly important in cardiac examinations where a high temporal resolution is essential to imaging the beating heart. Currently, the Somatom Sensation 64 and Definition (Siemens Medical) boasts the fastest gantry rotation speed of 330 milliseconds. Gantry rotation speed with a mounted x-ray tube due to gravitational forces may have nearly reached its limits.

Several other technical developments have occurred to give CT the place it has today, such as x-ray tube/generator technology, detector efficiency, and beam geometry, to name but a few; therefore, when comparing the products of several vendors, many technical factors have to be taken into consideration.

Current Vendor Options

Toshiba Medical?s Aquilion features high spatial resolution with the smallest detectors on the market.
The trademark of the Siemens Medical?s Somaton Definition is the dual-head scanner featuring two tubes at a 90° angle.

The development of CT has been accelerated by clinical need, although it is reasonable to say that a portion of clinical need has been realized through technological progress. A question that is asked by those involved in equipment-purchase decision-making is “Which CT scanner should I buy?” The answer to this question is not so straightforward and is obscure for good reasons. Perhaps more important than the choice of a specific vendor is how one optimizes the scan for the clinical question at hand, thus acknowledging the fluidity of imaging and the need to tailor acquisition parameters to a particular clinical question and even to a particular patient. Also important is the fostering of good relations with a vendor to enjoy a sneak preview of their future products and to build research collaboration. Vendors have a unique ethos, and this is translated into their final products; even as they attempt to outdo one another, it must be realized that an ideal CT scanner that encompasses the best features of all vendors is a legal impossibility, so a choice must, therefore, be made.

One might ask whether it is worth going from a 16-detector to a 64-detector scanner. The advantages of more detectors, such as faster scanning—thus more chance of a successful breath-hold or covering an extended field of view with a single contrast injection—have less incremental value with more detectors.

Sixteen detectors are probably enough for most applications, but not cardiac imaging, where the presence of additional detectors allows for multisegment reconstruction—a method to improve the temporal resolution—and permits faster scanning despite the low pitch required for retrospective gating. Additionally, the faster gantry rotation that has accompanied 64-detector CT is important for ensuring better temporal resolution. Indeed, the number of nonanalyzable coronary segments reported in the literature when using 16 detectors may be reason enough to warrant a 64 detector.

The following is a brief description of some of the unique features of the CT scanners offered by four major vendors. The features common to the vendors will not be elaborated upon.

GE Healthcare. Mark Morrison, marketing manager of CT, Americas, for GE Healthcare, succintly summarizes the company’s philosophy: “lower dose and better spatial resolution with no compromises in image quality.” Its VCT uses a fixed array of 64 detectors of individual width of 0.625 mm yielding z-axis coverage of 40 mm. VCT has high dose efficiency, as the focal spot matches the detector.

At the forefront of its cardiac package is Snap Shot Pulse, which allows prospective triggering of coronary artery imaging leading to considerable dose savings in patients with a regular heart rate. In prospective triggering, the tube current is on (and thus images are obtained) only in one phase (preselected to be the most motion-free) of the cardiac cycle. This technique gives anatomical information at lower doses than retrospective triggering, although no functional analysis is possible; this may not be a significant disadvantage if the primary purpose of CT is to assess anatomy, as function is easily assessed by echocardiography. However, the technique is best applied in those with a very low (<60 bpm) and stable heart rate.

VCT offers a higher limit for tube current/time product than is generally available from other vendors, a feature that may be employed occasionally when imaging morbidly obese patients.

CT brain perfusion is better enabled by a new platform known as Volume Shuttle. Here the table translates a precise distance of 40 mm, yielding a field of view of 80 mm that allows coverage of the circle of Willis and the neck vasculature.

Siemens Medical Solutions. The Somatom Sensation 64 and Definition are 64-slice, not 64-detector scanners. The detector configuration is an adaptive array design with the central 32 detectors of 0.6 mm and the outer 8 detectors of 1.2 mm. The longitudinal coverage in one gantry rotation is 19.2 mm if the central 32 detectors are used and 28.8 mm with the addition of the outer detectors. Z-sharp (focal flying spot) technology double samples the 32 detectors, producing effectively 64 slices of 0.4 mm at iso-center at any pitch. This method has the advantage of improving longitudinal resolution and reducing artifact that hitherto occurred because of aliasing (such as windmill artifacts).

The hallmark of Siemens Medical’s design—which underlies the company’s philosophy that more detectors may not be the way forward—is the dual-head scanner, Somatom Definition. By using two tubes at 90? to one another, data from only one quarter of a gantry rotation is required. This yields a temporal resolution of 83 m/sec, the highest in the industry, thereby allowing imaging of coronary arteries without the need for heart rate-lowering medications. In our department, this has tremendously improved patient throughput.

Philips Medical Systems. The hallmark of Philips Medical’s technology is user friendliness at both the scanner and workstation. The Philips Brilliance Scanner is a 64 detector with 40 mm of longitudinal coverage. The rate-responsive CVCT is patented software that permits phase registration with varying heart rates. Simply put, prescribing a particular cardiac phase either in terms of percentage of the R-R interval or as a temporal offshoot of the R wave assumes regularity of heart rate and any variation of the rate leads to phase mis-registration. The software allows better registration by using an algorithm that keeps systolic time relatively fixed and the diastolic time more variable in a patient with a variable heart rate.

Extended axial coverage is provided by the Jog Mode. Rapid View Reconstruction employs an advanced cone-beam reconstruction algorithm (COBRA) to provide rapid reconstructions, clearly valuable in both cardiac and trauma imaging.

Toshiba America Medical Systems. Not to be outdone by its competitors, the Aquilion 64 scanner boasts a high spatial resolution and its 64 detectors have a width of 0.5 mm—the smallest detectors currently in the market—resulting in a reconstructed resolution of 0.35 mm. The cone-beam reconstruction algorithms are advanced and based on Feldkamp principles. The SURE workflow offers an organ-specific postprocessing.

Future of CT

CT is evolving into a functional modality with greater anatomic clarity. Here are some areas to look out for:

  • More Detectors. The addition of detectors along the axial plane negated the need for axial translation (from second- to third-generation scanners); similarly enough, detectors can provide z-coverage to negate translation in the z-axis. In this case, we would have come a full spiral and back to axial imaging with the potential of covering an organ in a single gantry rotation, enabling perfusion imaging with dose savings. Toshiba’s 256 MDCT holds such a promise with preliminary results showing tremendous imaging quality and dose reduction in coronary imaging (coronary calcium score, angiogram, and perfusion imaging in one setting). More detectors would allow imaging without the assumptions of helical scanning (and without stair step artifact) but with greater cone beam geometry to reconcile. The benefits are not limited to cardiovascular imaging. Angiogenesis (and therefore perfusion) is a marker of tumor behavior and response to therapy. Serial dynamic perfusion imaging of neoplasm is likely to provide information regarding prognosis and therapeutic efficacy.
  • Flat-Panel Detectors. The move from solid-state to flat-panel area detectors (CsI-aSi) is an exciting avenue pursued by both GE and Siemens. Flat-panel detectors boast very high spatial resolution (of the order of 150 ?m), permitting evaluation of bone trabeculae and lung parenchyma with the clarity only thought to be available ex vivo. This is coupled with a large volume of coverage. One problem with flat-panel detectors is the afterglow, or the dead time, meaning that there is a limitation on temporal resolution (or gantry rotation time). The high spatial resolution will result in noise, which will need a higher dose to combat and contrast resolution is limited. Additionally, the large number of images will represent a computational challenge to store and transmit. For these reasons, flat-panel technology is unlikely to represent the next generation of scanners.
  • Fusion-Imaging. PET/CT is a symbiosis, where CT adds sensitivity to PET and PET adds specificity to CT, aiding in the detection of metastases and inflammatory foci and the assessment of myocardial viability, and ensuring targeted biopsy and radiation therapy. Fusion of images has been a challenge for non-neurological applications. The combined scanners provide a hardware approach to fusion with less mis-registration than seen with attempted fusion at the software level.
  • Dual-energy CT. Differentiating tissue even further based on the absorption spectra of the transmitted x-rays can be done by using two x-ray tubes with beams of different energies (Somatom Definition, Siemens Medical Solutions) or at the detector level. The latter strategy is being pursued by Philips and involves detectors that respond to a certain energy range. Siemens has recently obtained FDA approval for six applications of dual-energy CT, including “virtual noncontrast” imaging, plaque characterization, pulmonary perfusion, bone segmentation, and characterization of liver.

Many clinical challenges remain that will drive vendors to outdo one another in the CT race. This will take place against the backdrop of a quest to provide ever-less radiation. The imager of the future will view CT as a tool for virtual pathology, detailed anatomy, and confirmation of physiology, enabling the detection of disease at a far earlier and hopefully curable stage than is possible today.

Saurabh Jha, MD, is a fellow in cardiovascular imaging, and Harold Litt, MD, PhD, is chief of cardiovascular imaging in the department of radiology at the Hospital of the University of Pennsylvania, Philadelphia. For more information, contact .